Prospective evaluation of a model of prediction of invasive bacterial infection risk among children with cancer, fever, and neutropenia.

Abstract

A risk prediction model for invasive bacterial infection (IBI) was prospectively evaluated among children presenting with cancer, fever, and neutropenia. The model incorporated assessment of 5 previously identified risk factors: serum level of C-reactive protein (CRP) >/=90 mg/L, hypotension, identification of relapse of leukemia as the cancer type, platelet count of </=50,000 platelets/mm(3), and recent receipt of chemotherapy [16]. Children were uniformly evaluated at enrollment and were classified as having high or low risk for IBI according to a model that considers the number and type of variables present. Of the 263 febrile episodes evaluated during a 17-month period, 140 (53%) were in IBI-positive children. The sensitivity, specificity, and positive and negative predictive values of the model were 92%, 76%, 82%, and 90%, respectively. Identification of these 5 risk factors during the first 24 h of hospitalization was helpful in discriminating between children with a high or low risk for IBI.